action pattern

gistfile1.py

patterns = {
    0 :[
        # Common Pattern
        # click ad behavior
        (r'utm_source', r'ad'),
        (r'utm_source=([\w]+)', r'ad:\1'),
    ],
    28: [
        # Mayuki Pattern
        (r'mayuki', r'visit'),
        (r'showProduct', r'product'),
        (r'CrShopCar',   r'cart'),
        (r'CrShopCar03', r'payment'),
    ]
}
import re
def parse_actions(advertiser_id, url):
    actions = []
    ps = patterns.get(advertiser_id, []) + patterns[0]

    for pattern, repl in ps:
        if re.findall(pattern, url):
            actions.append(re.sub(pattern, repl, url))

    return actions

assert parse_actions(28, 'http://www.mayuki.com.tw') == ["visit"]
assert parse_actions(28, 'www.mayuki.com.tw/showProduct=123') == ["visit", "product"]
assert parse_actions(28, 'www.mayuki.com.tw/?utm_source=yahoo') == ["ad", "ad:yahoo"]